AI Agents Battle for Shopping Research
Stuart Kearney, co-founder of Vetted, on AI agents in shopping
This battle is really about who becomes the default place where a shopper decides what to buy before any merchant, marketplace, or checkout flow gets involved. Perplexity and OpenAI are moving upstream from search into product selection, where the winner gets the highest leverage point in commerce. That layer decides which brands make the shortlist, which sources count as trustworthy, and where purchase intent gets routed next. Vetted is built for that exact step, while Perplexity and OpenAI are trying to add it inside broader AI assistants.
-
The research layer is the messy part of shopping. People are comparing variants, price, delivery speed, returns, and whether a merchant is trustworthy. That is why Vetted focuses on narrowing choices and sending high intent traffic out, instead of trying to own checkout first.
-
Perplexity pushed further into native commerce earlier, with Shop like a Pro and Buy with Pro in November 2024. That gave it a way to turn answers into one click checkout, but only for supported merchants and products, which shows how hard it is to make universal shopping work reliably.
-
OpenAI is taking a similar path from a larger distribution base. ChatGPT now has a dedicated shopping research mode and has said direct purchase inside ChatGPT is coming through Instant Checkout. That means model companies are no longer just answering shopping questions, they are trying to capture transaction flow too.
The next phase is a split market between companies that own consumer intent and companies that own transaction rails. The strongest positions will belong to products that either become the shopping habit itself, like a specialized advisor, or plug directly into payment and checkout infrastructure that agents can trust at scale.